University of Groningen
Bistability, Epigenetics, and Bet-Hedging in Bacteria
Veening, Jan-Willem; Smits, Wiep Klaas; Kuipers, Oscar
Published in:
Annual Review of Microbiology
DOI:
10.1146/annurev.micro.62.081307.163002
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to
cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2008
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Veening, J-W., Smits, W. K., & Kuipers, O. P. (2008). Bistability, Epigenetics, and Bet-Hedging in Bacteria.
Annual Review of Microbiology, 62(1), 193-210. DOI: 10.1146/annurev.micro.62.081307.163002
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the
author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the
number of authors shown on this cover page is limited to 10 maximum.
Download date: 18-06-2017
ANRV354-MI62-11
ARI
ANNUAL
REVIEWS
5 August 2008
18:1
Further
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
Click here for quick links to
Annual Reviews content online,
including:
• Other articles in this volume
• Top cited articles
• Top downloaded articles
• Our comprehensive search
Bistability, Epigenetics,
and Bet-Hedging in Bacteria
Jan-Willem Veening,1,3 Wiep Klaas Smits,2,3
and Oscar P. Kuipers3
1
Institute for Cell and Molecular Biosciences, Newcastle University, Newcastle upon
Tyne NE2 4HH, United Kingdom; email: [email protected]
2
Department of Biology, Massachusetts Institute of Technology, Cambridge,
Massachusetts 02139; email: [email protected]
3
Molecular Genetics Group, Groningen Biomolecular Sciences and Biotechnology Institute,
University of Groningen, 9751 NN Haren, The Netherlands; email: [email protected]
Annu. Rev. Microbiol. 2008. 62:193–210
Key Words
First published online as a Review in Advance on
June 6, 2008
Bacillus subtilis, competence, sporulation, AND gate, phenotypic
variation, synthetic biology
The Annual Review of Microbiology is online at
micro.annualreviews.org
This article’s doi:
10.1146/annurev.micro.62.081307.163002
c 2008 by Annual Reviews.
Copyright All rights reserved
0066-4227/08/1013-0193$20.00
Abstract
Clonal populations of microbial cells often show a high degree of phenotypic variability under homogeneous conditions. Stochastic fluctuations in the cellular components that determine cellular states can cause
two distinct subpopulations, a property called bistability. Phenotypic
heterogeneity can be readily obtained by interlinking multiple gene
regulatory pathways, effectively resulting in a genetic logic-AND gate.
Although switching between states can occur within the cells’ lifetime,
cells can also pass their cellular state over to the next generation by a
mechanism known as epigenetic inheritance and thus perpetuate the
phenotypic state. Importantly, heterogeneous populations can demonstrate increased fitness compared with homogeneous populations. This
suggests that microbial cells employ bet-hedging strategies to maximize survival. Here, we discuss the possible roles of interlinked bistable
networks, epigenetic inheritance, and bet-hedging in bacteria.
193
ANRV354-MI62-11
ARI
5 August 2008
18:1
Contents
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
PHENOTYPIC VARIATION
AND ITS ORIGINS . . . . . . . . . . . . . .
NETWORK TOPOLOGY . . . . . . . . . . .
Noise, Hysteresis, and Bistability . . .
COMPETENCE FOR GENETIC
TRANSFORMATION
IN BACILLUS SUBTILIS . . . . . . . . . .
PROSPECTS OF USING BISTABLE
SWITCHES FOR
BIOTECHNOLOGY AND
SYNTHETIC BIOLOGY . . . . . . . . .
CELL AGE AND ITS ROLE IN
PHENOTYPIC VARIATION . . . . .
EPIGENETIC INHERITANCE OF
PHENOTYPIC VARIATION . . . . .
Memory Within the lac Operon . . . .
Sporulation in B. subtilis . . . . . . . . . . . .
GENETIC LOGIC-AND GATES . . . .
Heterogeneity in Exoprotease and
Biofilm Matrix Production . . . . . .
Hypermutable Subpopulations
in E. coli . . . . . . . . . . . . . . . . . . . . . . . .
PHENOTYPIC VARIATION AS A
BET-HEDGING STRATEGY . . . .
Bacterial Persistence . . . . . . . . . . . . . . .
Sporulation Bistability as a
Bet-Hedging Strategy . . . . . . . . . . .
OUTLOOK. . . . . . . . . . . . . . . . . . . . . . . . . .
194
194
194
196
198
198
198
199
200
202
202
203
204
204
205
205
PHENOTYPIC VARIATION
AND ITS ORIGINS
lacZ: β-galactosidase,
traditional reporter
that cleaves colorless
X-gal, resulting in
bright blue products
Bet-hedging: a risk
spreading strategy to
diversify phenotypes
with the aim to
increase fitness in
temporally variable
conditions
194
Bacterial growth is traditionally viewed as the
result of (symmetrical) cell division yielding
siblings that are genetically identical. Consequently, the results from reporter studies such
as those employing lacZ have traditionally been
interpreted using the assumption that all cells in
a culture behave in an identical manner. However, it has long been recognized that within
isogenic populations, bacterial cells can display various phenotypes. This microbial cell
individuality or phenotypic variation is receiving increased attention because of its releVeening
· ·
Smits
Kuipers
vance for cellular differentiation and implications for the treatment of bacterial infections
(92). Phenotypic variation is a widespread phenomenon in the bacterial realm. Some of the
well-characterized examples include the lysislysogeny switch of phage lambda, lactose utilization and chemotaxis in Escherichia coli, phase
variation in a number of pathogens, and cellular differentiation in Bacillus subtilis (for recent
reviews see References 11, 25, and 92). Strikingly, many documented cases of phenotypic
variability relate to responses to environmental
stresses, suggesting that phenotypic variation
aids in the survival of cells under adverse conditions and therefore may be an evolvable trait.
The potential function of phenotypic variation
as a bet-hedging strategy is further elaborated
upon in other parts of this review.
Various different mechanisms are involved
in phenotypic variation. Phenotypic differences
can be due to mutation, variations in the microenvironment, mutation, phase variation, cell
cycle, and the wiring of the network that governs a specific stress response (11, 92). The focus of this review is on the role of phenotypic
variability that results from amplified noise in
gene expression.
NETWORK TOPOLOGY
As early as 1961, Monod and Jacob postulated
that the differences in the response of individual
cells to a stimulus could in theory be explained
by the architecture of the underlying gene regulatory network (66). However, their hypotheses could not be experimentally addressed until the development of single-cell techniques
and were not computationally tractable until recently. Considering the importance of this type
of mechanism in generating phenotypic variation (92), it is discussed in more detail below.
Noise, Hysteresis, and Bistability
In biological systems, signals are never discrete
because of random fluctuations in the biochemical reactions in the cell. This stochastic variation is called noise and is a key determinant
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
ANRV354-MI62-11
ARI
5 August 2008
18:1
of phenotypic variation (49, 81, 85). Noise is
predicted to be most dominant when the number of molecules involved is small (finite number effect). Experimental verification of this notion came from fluorescent reporter studies (28,
78, 98). This effect is notable for two reasons.
First, transcription and translation are thought
to generally involve relatively small numbers
of molecules compared with, for instance, the
numbers of molecules participating in proteinprotein interactions. Second, when not activated, transcription factors are usually in low
abundance. Moreover, many stress responses
are accompanied by a reduction in general transcriptional and/or translational efficiency (38).
This potentially leads to an induction of phenotypic variation under these conditions. Generating variable phenotypes may be beneficial for
the survival of populations under adverse conditions, and stimulating noisy expression might
be an elegant way of achieving this (72).
Noise can be exploited under certain conditions to generate phenotypic heterogeneity.
For example, noise in the regulatory cascade
that governs the chemotactic response of E. coli
results in behavioral individuality with respect
to the rotational direction of the flagella (54).
When a noisy signal is amplified by net positive
feedback, gene expression levels can be further
bifurcated and this situation deserves special attention. In the presence of positive feedback, a
graded response (i.e., with intermediate levels
of expression) can be converted to a binary response, in which cells express a certain gene at
high or low levels (13). At the population level,
this switch-like behavior can result in a bimodal
distribution in gene expression because some
cells switch, whereas others do not. This type of
gene expression pattern is commonly referred
to as bistability (25, 92).
In physics, multistationarity describes a network that has more than one stable state. Extending this to biology, it means that a gene
regulatory network potentially exhibits two (or
more) discrete levels of gene expression (a high
state and a low state). Bistability describes a
parameter regime in which a dynamic system
can rest in either of two stable states. Anal-
ogous to the previous definition, it refers to
conditions under which cells can be in a highexpressing or low-expressing state for biological
systems. Multistationarity at the cellular level is
an intuitive explanation for population bistability; hence, the terms are frequently used interchangeably. Although most biological systems
that demonstrate population bistability involve
noise amplified by some form of net positive
feedback, they are not necessarily bistable in a
deterministic sense (95).
The requirements for a gene network to exhibit multistationarity have been explored in
detail (29, 92). In summary, the system needs
to display nonlinear kinetics in addition to positive feedback. For transcriptional regulators,
nonlinearity can be the result of multimerization, cooperative binding to target sequences on
the DNA, or phosphorylation of certain amino
acid residues. In many cases nonlinearity is evident as a sharp increase in the expression of a
downstream target gene above a certain threshold level of the regulator. Only networks that
include an even number of negative-feedback
loops and/or any number of positive-feedback
loops are capable of causing multistationarity
(8). Experimentally, some bistable gene expression patterns rely on positive feedback as well
as double-negative feedback (toggle switch)
(92 and references therein). However, positive
feedback in itself is no guarantee for bistability
(29), and bistability is also possible when based
on other types of network architecture (8) or
mechanisms such as multisite phosphorylation
(55, 77).
A common feature of bistability is hysteresis (74). Hysteresis refers to the situation in
which the transition from one state to the other
requires an induction (or relief of induction)
greater than that for the reverse transition.
This imposes memory-like characteristics onto
the network (see also Epigenetic Inheritance of
Phenotypic Variation, below), making the response of cells dependent on their recent history. Hysteresis in biological systems can reside,
for instance, in the stability of one of the proteins involved. When Novick & Weiner (76)
described the all-or-none enzyme induction
www.annualreviews.org • Bistability in Bacteria
Multistationarity:
multiple stationary
stable states within a
(genetic) network
between which
switching is possible
Bistable: a network
with two steady states,
or two distinguishable
phenotypes within a
clonal population
195
ANRV354-MI62-11
ARI
5 August 2008
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
Sporulation: a
developmental process
ultimately resulting in
the formation of a
highly resistant
(endo)spore
Competence: the
ability to take up DNA
from the environment
and stably maintain its
information in the
genome
18:1
in lactose utilization they noted that at nearthreshold concentrations of inducer the population of E. coli cells segregated into two subpopulations, which is now regarded as one of
the earliest examples of bistability. Subsequent
experiments revealed that the history of the
inoculum influenced the fraction of cells in
each subpopulation (23). The hysteretic behavior of the multistable lactose utilization network is a result of the stability and abundance
of the lactose permease (79, 107). Hysteresis
can act as a buffer, reducing accidental switching between states due to minor perturbations
(1, 16).
Although bistable systems are in principle
reversible, the time required for a cell to revert to the initial state (escape time) may exceed the duration of the experiment or even
Direct effects
Indirect effects
ComK
Positive feedback
ComK
Toggle switch
Excitable
Rok
ComS
Figure 1
Regulatory elements featured in current models on competence development
in Bacillus subtilis. A highly simplified representation of the three core elements
of the competence regulatory network: (a) ComK autostimulation is
responsible for a positive-feedback loop required for a bimodal expression
pattern, (b) a putative toggle switch is dispensable for a bimodal expression
pattern, and (c) interlinked positive- and negative-feedback loops that result in
excitable behavior are implicated in the temporal nature of competence. Solid
lines represent direct or well-characterized interactions; dotted lines represent
putative or indirect effects.
196
Veening
· ·
Smits
Kuipers
the lifetime of the organism. Moreover, phenotypic switches can be rendered unidirectional by
downstream signaling events. For instance, the
bistable switch governing sporulation in B. subtilis becomes irreversible after its earliest stages
owing to an orchestrated sequence of events
(26).
COMPETENCE FOR
GENETIC TRANSFORMATION
IN BACILLUS SUBTILIS
To further explore general mechanisms by
which phenotypic variation can arise, we discuss
one of the best-understood naturally occurring
bistable systems in bacteria: competence development in B. subtilis. The first evidence for
the existence of subpopulations in a competent
culture of B. subtilis came from elegant experiments that demonstrate biosynthetic latency of
competent cells (17, 41, 73). Subsequently, the
expression of the key regulator of competence
development, ComK, was limited to the competent fraction of the culture (42).
ComK is a multimeric transcription factor
that is necessary and sufficient to activate the
expression of all genes that encode the DNA
uptake and integration machinery by binding
to a consensus motif in the target promoters
(44). Key features of the complex regulatory
network that controls ComK levels are transcriptional regulation at the comK promoter
and proteolytic degradation of ComK protein
(44). ComK stimulates its own expression by
reversing the effects of at least two repressors,
one of them named Rok (for repressor of
comK ), establishing a positive-feedback loop
(91). Additionally, ComK is believed to repress
transcription of rok. This interaction forms a
putative toggle switch. Proteolytic degradation
of ComK is antagonized by the anti-adaptor
protein ComS, which is required for the initiation of competence. Evidence suggests an indirect negative-feedback loop, as overproduction
of ComK inhibits ComS expression (95). The
features described above are summarized in
Figure 1, and they all form modules that are
potentially involved in phenotypic variation.
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
ANRV354-MI62-11
ARI
5 August 2008
18:1
ComK autostimulation is necessary and can
be sufficient to establish a bimodal expression
pattern (61, 90), but it is independent of Rok,
excluding a toggle switch–like mechanism. The
transition between low- and high-expressing
states was attributed to stochastic fluctuations
in conjunction with the positive-feedback loop
that would amplify the signal as the concentrations of ComK exceed a certain threshold
(103). The role of noise was experimentally
addressed in two studies. Süel and coworkers
averaged out the noise of multiple cells by depleting cells for ftsW, which is required for septation. An analysis of competence development
under those conditions revealed that the chance
of initiation of competence was greatly reduced
(96). In a more direct approach, Maamar and
coworkers (62) adopted a method derived from
Elowitz et al. (28) to show that intrinsic noise
in comK expression selects cells for competence.
Reducing intrinsic noise, by increasing transcriptional efficiency and reducing translational
efficiency, caused significantly less cells to enter competence. Their findings are consistent
with another report that demonstrated significant variation in basal promoter activity of
comK (57). Because ComK is responsible for the
activation of the late competence genes (such
as comG), intrinsic noise in comK expression
results in pathway-specific extrinsic noise in
comG expression.
Competence is a transient process; under
laboratory conditions it is limited to several
hours in stationary growth phase or until cells
are resuspended in fresh growth medium. Although the molecular mechanisms responsible
for escape from the competent state remain elusive, mathematical modeling has recently shed
some light on potential mechanisms and has led
to the development of two predominant models. Both models share the notion that noise is
amplified by the ComK autostimulatory loop.
In the bistable model, intrinsic noise of comK
expression (57, 62) is critical for the switching
of cells from the noncompetent to the competent state. In the excitable model, the source
of the noise that triggers the excursion from
the vegetative state remains undefined (95, 96).
Although both models can result in a bimodal
distribution at the population level (as both involve stochastic switching), only the first model
is bistable in a deterministic sense. The excitable model generates a bimodal gene expression pattern because the transition to the highexpressing state is fast compared with the slowly
acting negative-feedback loop, but the high expression level does not represent a stable state
(95).
Both models offer a different explanation
for the temporal nature of competence. In the
bistable model, two mechanisms are at play.
First, cells can revert from the high-expressing
to the vegetative state by stochastic transitions.
Second, the basal promoter activity of comK, as
measured by the number of mRNA molecules
per cell, is greatly reduced in stationary growth
phase (57, 62). This causes a window of opportunity for cells to switch to the competent state
and generates conditions under which the saturated proteolytic complex reduces ComK levels
enough to escape the competent state. The validity of this hypothesis was confirmed through
mathematical modeling (62).
The excitable model offers an attractive hypothesis for the limited time span during which
cells are competent for DNA uptake. In contrast to the bistable model, the competent state
is not stable owing to the action of a slowly acting negative-feedback loop. As a result cells will
always return to the vegetative and stable state.
The model makes some predictions about the
dynamics of the competence network that are
experimentally addressed using time-lapse fluorescent microscopy (96).
The elegance of the excitable model has attracted a lot of attention, as it resembles the
dynamics of oscillatory systems such as cell cycle and circadian rhythms. However, it fails to
couple back to the observations made in singlecell analyses of competent cultures that demonstrate a limited time frame during which competence occurs in a culture and does not take
the observed decrease in basal comK transcription into account. Although certain features of
www.annualreviews.org • Bistability in Bacteria
Excitable: a transient
excursion from a stable
state leading to the
expression of a
phenotype in a limited
period of time
197
ANRV354-MI62-11
ARI
5 August 2008
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
Epigenetic
inheritance (EI):
inheritance of a
phenotype from one
generation to the next
that does not depend
on changes in DNA
sequence
18:1
the two models are not reconcilable, it is possible that both mechanisms occur in nature under
different conditions, for example, the timescale
on which they occur could vary. Moreover, it
has been suggested that stochastic activation
of comK in combination with positive feedback
could result in a bimodal expression pattern,
even in the absence of bistability in the deterministic sense (9, 50). It is a challenge for future investigators to address these unanswered
questions.
PROSPECTS OF USING
BISTABLE SWITCHES FOR
BIOTECHNOLOGY AND
SYNTHETIC BIOLOGY
Construction of synthetic genetic circuits using naturally occurring cellular components
in living cells allows them to be tested separately from the context of other physiological
processes. Synthetic switches are operational
in prokaryotic and mammalian cells and valuable for gaining insight into naturally occurring genetic circuitries (45, 46). Synthetic biology also allows the creation of entirely new,
or rerouted, networks, such as toggle switches,
oscillatory networks, and even synthetic multicellular clocks based on quorum sensing (10,
27, 34, 36). Some of these findings made it to
patents (37), showing the realistic prospect of
industrial utilization of engineered circuitries
leading to phenotypic variation.
Combinatorial promoter design also is effective for engineering noisy gene expression
(71), and various successful examples of combinatorial promoter design have been published
(24, 43). Global transcription machinery engineering (gTME) is a compatible strategy for
improving metabolic engineering efforts. Instead of direct enzyme or metabolic pathway
engineering, gTME reprograms the transcription machinery, resulting for example in increased ethanol tolerance and production in
yeast (7). This method could be well combined
with the strategies outlined above to engineer
novel regulatory circuits.
198
Veening
· ·
Smits
Kuipers
CELL AGE AND ITS ROLE IN
PHENOTYPIC VARIATION
Although aging has already been described
to cause phenotypic variability in yeast (4),
Caulobacter crescentus was the first bacterium for
which aging was demonstrated (3). It was found
that the reproductive output of cells decreased
with age. Asymmetric division is a hallmark
of the life cycle of this bacterium, and these
observations are therefore consistent with the
hypothesis that mortality requires asymmetry
(80).
In many other prokaryotes, however, cell division leads to two visibly identical daughter
cells, and as a result, they have been regarded as
nonsenescent. Yet, the subcellular localization
of a set of proteins may distinguish old and new
poles in morphologically symmetrical bacteria.
By following single E. coli cells through several
rounds of cell division, Stewart and coworkers
showed that growth rate inversely correlates to
cell pole age, demonstrating that aging is not
limited to organisms with asymmetric division
(94). It was recently found that aggregated proteins and chaperones preferentially accumulate
at the old cell pole (59), reminiscent of the situation in yeast in which oxidatively damaged
proteins accumulate in the mother cell (4).
Recently, time-lapse microscopy has been
used to follow the growth, division, and cellular differentiation of individual cells of B. subtilis (104), an organism that is well known for
asymmetric division prior to the formation of
an endospore. The study revealed that B. subtilis, like E. coli and C. crescentus, suffers from
aging but that spore formation is not biased toward either the old or the new cell pole (104).
Interestingly, the magnitude of this aging effect is nearly identical to that seen in E. coli and
C. crescentus.
EPIGENETIC INHERITANCE
OF PHENOTYPIC VARIATION
Epigenetic inheritance (EI) (or non-Mendelian
inheritance) is the passage of cellular states from
one generation to the next, without alterations
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
ANRV354-MI62-11
ARI
5 August 2008
18:1
of the genome (48). The classic example of EI
is the stable transfer of a phenotype by modifications to the DNA such as methylation (19).
This modification can be stable over multiple
rounds of cell division but it does not involve
actual changes in the DNA sequence of the organism. Other epigenetic phenomena include
prions, genomic imprinting, and histone modification (19 and references therein).
It has been proposed that autophosphorylating kinases have the potential to store memory.
In this scenario, a specific stimulus activates the
kinase, and because of its autocatalytic properties the kinase stays in its active state, regardless
of the presence or absence of the stimulus (60).
As a result, the progeny of cells in the ONstate will also be in the ON-state because the
activated kinase is passed on to the offspring.
Using artificial bistable gene regulatory circuits
in both E. coli and Saccharomyces cerevisiae, autostimulatory regulation systems can function
as memory devices in microorganisms (13, 36).
EI of phenotypic variation can also be based
on the transfer of active transcriptional regulators during cell division via positive feedback
(18, 60, 84). When cells divide, not only DNA
but also cellular factors such as proteins and
RNA are partitioned, and importantly, this can
dictate future life-history decisions of the new
offspring. Valuable knowledge on the molecular mechanism responsible for EI and the minimal requirements to generate stable inheritance
of phenotypic variation is arising from studies
using well-defined artificial gene networks (6,
51). The simplest network that demonstrates EI
is one in which a positive regulator autostimulates its own promoter upon stimulation by an
exogenous signal. Once activated, the positive
feedback of the system will ensure high intracellular levels of the positive regulator, regardless
of the absence or presence of the signal. In such
a system, the degradation rate of the regulator
and the growth rate of the cell are determining
factors of the stability of the memory response
(6).
An example of a simple (but general) network motif that putatively generates EI is depicted in Figure 2. A number of requirements
need to be met before EI can occur. The network should show two stable steady states (activator OFF and activator ON). This depends
on activator production/decay rate and growth
rate, and activator production should be cooperative (6). In addition to this, the basal activator levels should be at a level lower than
required to autoactivate its own synthesis; otherwise cells will always be in the ON state. Furthermore, once the system is activated, activator levels should be high enough to drive its
own expression; if not, cells will quickly switch
back from the ON to the OFF state and EI
cannot be established. Even cell fates driven
by a semistable stochastic switch with reduced
positive feedback inherit epigenetically. This is
likely caused by initial bursts of activator protein in the mother cell, which maintains at high
levels through multiple rounds of division (51).
Two examples of the significance of EI of phenotypic variation in bacteria are discussed below. Other instances, primarily from eukaryotes, fall outside the scope of this review.
Memory Within the lac Operon
As discussed above, bistable systems depend on
some form of positive feedback within the gene
network. The first epigenetic system described
in bacteria is the lac operon of E. coli (76). The
genes that encode the proteins required for the
uptake and utilization of lactose are induced
in the presence of the gratuitous (nonmetabolizable) lactose analogue, isopropyl-d-thio-βgalactopyranoside (IPTG). At high IPTG concentrations the lac operon is fully derepressed
and cells highly express the IPTG permease
protein and thus remain highly activated. At low
concentrations, however, cells that were previously uninduced and do not have any permease in their membranes do not respond to
the low level of IPTG and remain in the OFF
state. Cells that were previously induced and
still have some permease are activated by the
low level of IPTG and remain in the ON state.
Reculturing of single cells results in populations that either give high or low lac expression
(70 and references therein). This phenomenon
www.annualreviews.org • Bistability in Bacteria
199
ANRV354-MI62-11
ARI
5 August 2008
18:1
Activator
Inactive postive
feedback
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
Activator
mRNA
Stimulus
Cell in
ON state
Figure 2
Epigenetic inheritance by positive feedback. A basal level of activator protein and mRNA (single helix) is always present regardless of the
absence of stimulus (lightning symbol ). However, this basal level is insufficient to activate the positive-feedback loop (red X) and activator
protein levels remain low. When the signal is present, however (which might be caused by noise), activator protein multimerizes and
stimulates its own expression, resulting in high concentrations of activator, and in this example, high activator concentrations induce multimerization. Because of the positive-feedback loop, intracellular activator concentrations remain above the threshold required to stimulate
transcription and cells remain in the ON state ( green cells) for multiple generations even in the absence of stimulus. Cell growth and division
can dilute activator, but as long as the concentrations remain high enough to drive promoter firing, cells will remain in the ON state.
is called all-or-none enzyme induction (76) and
is indicative of the presence of two coexisting
subpopulations. The permease plays a pivotal
role and constitutes the positive-feedback loop
in this system: High permease levels keep the
levels of intracellular IPTG high, thus inducing
permease gene expression. Importantly, under
low inducer conditions, either the ON or OFF
state can be epigenetically inherited by the offspring through multiple rounds of growth and
division. In this situation, the physiological state
of the offspring is a reflection of the past state
of its ancestor. A possible explanation for such a
positive-feedback loop in the lac operon is that
in the presence of (metabolizable) lactose, the
IPTG: isopropyld-thio-βgalactopyranoside
200
Veening
· ·
Smits
Kuipers
E. coli population can quickly drain the sugar
pool even when the sugar concentration starts
to decrease (18).
Sporulation in B. subtilis
Sporulation of B. subtilis has been described
as a bistable process because two distinct subpopulations can be distinguished within an
isogenic population of stationary-phase cells:
sporulating and nonsporulating cells (reviewed
in References 25 and 92). Initiation of sporulation is driven by the master sporulation regulator Spo0A. A basal level of Spo0A is always present, and upon specific environmental
ARI
5 August 2008
18:1
signals such as high cell density and nutrient deprivation, Spo0A is phosphorylated and directly
activates expression of more than 100 genes, including its own gene (31, 65). Sporulation bistability is not a simple ON/OFF switch, because
the levels of Spo0A∼P increase gradually after
activation (32). Recent research has shown that
although the positive feedback of Spo0A∼P on
spo0A transcription plays an important role in
the distribution of cellular states (31, 101), it
is not critical in establishing sporulation bistability (104). Rather it seems that the activity of
the phosphorelay dictates sporulation bistability because cells constructed to express a mutant form of Spo0A (Sad67) (47) that does not
require activation no longer show bistability
(104).
A recent study using time-lapse microscopy
found a strong correlation between cell lineage
and the decision to sporulate or not sporulate
(104). Close relatives often demonstrate a similar phenotype (to either sporulate or not sporulate). Phylogenetic reconstruction of sporu-
a
lating microcolonies using parsimony analyses
showed that the decision to sporulate could
often be traced back more than two generations before the actual appearance of the phenotype (Figure 3). This finding indicates that
the signal to sporulate already occurs during
the logarithmic growth phase and is epigenetically passed on. Again, an important role for the
sporulation phosphorelay was identified for this
epigenetic effect (104), indicating that bistability is a prerequisite for EI of the sporulation
signal.
The putative benefits of EI within a sporulating population are complex. For cold-shock
adaptation in bacteria, cells pretreated by a mild
cold shock memorize this stress and are better prepared for a harsher cold shock, which
would otherwise be lethal (40). In analogy to
this, it can be envisaged that propagation of the
sporulation signal from the mother cell to its
descendants helps the progeny to be prepared
for potential nutritional limitations in the future in such a way that they can rapidly respond
b
Cells ON
Cells OFF
Cells OFF
Cells ON
Genealogical position
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
ANRV354-MI62-11
0
200
400
600
1
1-5
1-4
1-4-1
1-3
1-3-2
1-3-1
1-3-1-1
1-2
1-2-1
1-2-1-1
1-1
1-1-2
1-1-2-1
1-1-1
1-1-1-1
1-1-1-1-1
800
Time (min)
Figure 3
Lineage reconstruction to plot cell fate distributions within isogenic populations. (a) Parsimony
reconstruction of the sporulation signal within a Bacillus subtilis microcolony. Every node in the radial tree
represents one cell division event. Every endpoint in the tree represents one offspring cell. Orange tips are
cells that have activated Spo0A. Parsimony reconstruction shows the first appearance of a mother cell that
creates offspring of mostly cells in the ON state (orange lines). Figure from Reference 104. (b) Family tree of
Saccharomyces cerevisiae harboring an artificial bistable switch. Gray lines indicate cells in the OFF state,
whereas orange lines represent cells after they have switched to the ON state. In this genealogy graph, in
contrast to panel a, line length is a direct measure of time. Figure from Reference 51.
www.annualreviews.org • Bistability in Bacteria
201
ANRV354-MI62-11
ARI
5 August 2008
CRIF: cis-regulatory
input function
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
Altruism: a behavior
that decreases the
fitness of the altruistic
individual while
benefiting others
18:1
and commit to spore formation when required.
Alternatively, EI may serve to coordinate multicellular behavior (104), a process which in
B. subtilis is also dictated by Spo0A (5).
GENETIC LOGIC-AND GATES
Often, transcription of a gene is regulated by
more than one regulator (input). The way these
inputs control the transcription rate (output)
is described by the cis-regulatory input function (CRIF) (64). CRIFs can often be described
by Boolean-type functions such as logic-AND
gates and logic-OR gates (64 and references
therein). Synthetic logic-AND gates can be exploited to program specific responses of cells
(75). If one of the inputs of a CRIF is heterogeneous and the target gene is under control of a
logic-AND gate, then by definition the output
is also heterogeneous.
A number of studies recognize that certain genes of one pathway are heterogeneously
expressed because their regulation is interlinked with another (bistable) network through
a logic-AND gate. The use of an AND gate
system is a simple strategy to generate phenotypic variability without the necessity to create complex switches with multiple steady states
(Figure 4a). Here we consider a few examples
in which heterogeneity in gene expression can
be ascribed to the logic of the underlying circuitry. We discuss the putative physiological
relevance of the observed heterogeneity as a result of the AND circuit.
Heterogeneity in Exoprotease
and Biofilm Matrix Production
Recently, it was found that high expression
of aprE (subtilisin) and bpr (bacillopeptidase),
two important extracellular proteases (exoproteases) of B. subtilis, is limited to only a small
part of the population (Figure 4b) (102). Exoprotease production has been described as a
survival strategy under nutrient-limiting conditions, and these enzymes act as scavenging proteins that degrade (large) proteins into smaller
fragments that can be subsequently taken up as
202
Veening
· ·
Smits
Kuipers
a new nutrient source (68). Studies using wild
B. subtilis strains also indicate a role for exoproteases during biofilm formation (53, 105).
Expression of both aprE and bpr is under
the control of the DegS-DegU two-component
system (68). To activate aprE gene expression,
DegU needs to be phosphorylated by the DegS
sensor protein (69). In addition, aprE is under direct negative control of at least three
other transcriptional regulators (AbrB, SinR,
and ScoC), all of which are under direct or
indirect negative control by the key sporulation regulator, Spo0A∼P (31). The result of
this intertwinement with the sporulation pathways is that aprE will only be derepressed in a
subpopulation when nutrients become limiting.
Together, the aprE gene regulatory network
acts as a logic-AND circuit in which a threshold level of dimerized DegU∼P and Spo0A∼P
is integrated to activate gene expression (102)
(Figure 4c).
It has been hypothesized that cells that produce and secrete these proteases help not only
themselves, but all clonal cells within the local
growth medium. This might be regarded as a
simple form of altruism. One explanation for
altruism is when the cooperation is directed toward individual cells that are genetically similar
(kin selection) (63). Heterogeneity in gene expression ensures that not all cells commence
into the costly production of Bpr and AprE,
but all cells within the clonal population benefit from the activity of these extracellular
proteases.
Similarly, the extracellular matrix within
biofilms of B. subtilis is produced by a small
fraction of cells within the population (20,
106). Expression of the products that form
the extracellular matrix (EpsA-O, YqxM, and
TasA) is under direct negative control of SinR,
the master biofilm regulator in B. subtilis (52).
This regulator is antagonized by SinI, a protein under control of Spo0A. sinI seems to
be activated by low levels of Spo0A∼P but
repressed at high levels of Spo0A∼P (20), although this still awaits experimental validation.
Thus, expression of sinI and, as a result, the
genes responsible for the extracellular matrix
ANRV354-MI62-11
ARI
5 August 2008
18:1
a
b
Input 1
Output
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
Input 2
c
Spo0A~P
aprE
Exoprotease production, biofilm formation
sinI
Extracellular matrix, biofilm formation
e.g.,
dinB
Hypermutable subpopulation
DegU~P
Spo0A
ON
Spo0A~P
low
d
DSB
SOS
RpoS
Figure 4
Naturally occurring genetic logic-AND gates. Arrows indicate positive actions. (a) Input 1 (red arrow) and
input 2 (blue arrow) are active only in ∼50% of the isogenic population (red and blue cells, respectively). Cells not
active for input 1 or 2 are depicted in white. The output of the system ( yellow arrow) is expressed only when both
input 1 and input 2 are active within the same cell ( yellow). As a result of this AND gate, the isogenic population
of four distinguishable phenotypes can exist: white, red, blue, or yellow cells. (b) Heterogeneous expression
of aprE. Expression of aprE is monitored by a fusion of the aprE promoter to gfp. Within stationary-phase
cultures, three distinct phenotypes can readily be observed: sporulating cells, vegetative cells, and aprE-gfpexpressing cells. (c) The genetic circuit responsible for aprE and biofilm heterogeneity. Thick arrows indicate
that the system can be overridden by (artificial) induction of the activator protein. For simplicity, we depict the
effect of Spo0A∼P on the multiple repressors that act on aprE as a single positive arrow. (d ) SOS response and
RpoS requirement for the formation of a hypermutable subpopulation in Escherichia coli. See text for details.
within biofilms are activated only when two
conditions are met: (a) Spo0A needs to be activated and (b) Spo0A∼P levels cannot be too
high. Although this is not a true logic-AND circuit, the result of the network wiring is that only
a small subpopulation of cells expresses sinI. Because the production of the extracellular matrix
within biofilms is energetically costly, the division of labor might enhance the total fitness of
the entire bacterial community. It will be interesting to see how this labor is divided in multispecies biofilms.
Hypermutable Subpopulations
in E. coli
Clonal populations of cells may diverge owing to changes in their genetic makeup. The
www.annualreviews.org • Bistability in Bacteria
203
ANRV354-MI62-11
ARI
5 August 2008
Adaptive
mutagenesis:
describes a set of
conditions under
which mutations
appear to occur more
often when selective
pressure is present
than when not
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
HMS: hypermutable
subpopulation
DSB: double-strand
break
Persistence: the
phenomenon of the
existence of a small
subpopulation of cells
that do not grow
compared with the rest
of the isogenic culture,
and as a result are
antibiotic resistant
18:1
occurrence of mutations may give certain cells
a selective advantage over others, and this may
cause a subpopulation to form or even take over
the culture (108).
Under conditions of stress, adaptive mutagenesis (and/or stationary-phase mutagenesis)
can occur (for recent review see Reference 30).
In E. coli, adaptive mutations were associated
with other, unselected mutations, indicating
the existence of a hypermutable subpopulation
(HMS) (100). The observed hypermutation is
not caused by a stable mutator phenotype that
could result from genetic differences, but reflects a transient differentiated state (39, 87).
The mechanisms involved in hypermutation include double-strand break (DSB) repair,
SOS response, and a general stress response, of
which the first two have a causal relationship
(33). The critical factor in HMS, though not
the only one, is that cells are continuously facing DNA double-strand breaks, even in the absence of external DNA-damaging agents. The
induction of DSB repair is evidenced by the formation of foci of RecA protein, a key protein
in the repair pathway, in a subset of cells (86,
88). DSBs can lead to induction of the SOS response, and ∼1% of growing E. coli cells is SOS
induced under steady-state conditions (82).
The switch to HMS requires an additional
requirement to be satisfied, as artificially induced DSBs do not lead to HMS until cells enter stationary phase (83). At that time, the levels of the general stress sigma factor RpoS rise,
and it was found that artificially inducing RpoS
can lead to HMS in exponential growth phase
(83). Thus, the preexisting heterogeneous input of (at least) the SOS response, together with
RpoS, forms a logic-AND gate that leads to the
formation of the HMS (Figure 4d ).
PHENOTYPIC VARIATION
AS A BET-HEDGING STRATEGY
A major question that arises from the finding
of population heterogeneity is, Why do bacteria display phenotypic variation? The most apparent hypothesis is that this strategy is a form
of bet-hedging. Under challenging conditions,
204
Veening
· ·
Smits
Kuipers
the production of offspring with variable phenotypes ensures that at least one offspring will
be appropriate (fit) under a given situation (22).
This is a risk-spreading or bet-hedging strategy,
because not every offspring will be optimally
suited for the future environment. However,
the overall fitness of the genotype will increase
because some offspring will have the proper
adaptation. Although heterogeneity might not
be ideal under homogenous, steady-state conditions, mathematical studies support the notion
that in a variable environment a heterogeneous
population outcompetes (or is fitter than) a homogeneous population (56, 99). Importantly, it
was suggested that phenotypic variation is an
evolvable trait. This was recently underscored
in an elegant study on S. cerevisiae, in which
interphenotype switching rates, like those between the two stable states of gene expression
in a bistable system, are tuned to the frequency
of changes in the environment (2).
Experimental evidence for the benefits of
phenotypic variation is limited. In yeast, clonal
populations with increased variability in stress
resistance are more successful than strains with
limited variability under conditions of stress
(15). Moreover, heterogeneous populations of
yeast outcompete homogenous populations under cadmium stress conditions (89).
Bacterial Persistence
Originally identified in 1944 (14) persistence is
one of the best-documented examples of a bacterial bet-hedging strategy (for a recent review
see 58). Persister cells are not simply antibiotic
resistant but rather reflect a transient growth
arrested state. Persister cells can be grown to
form a population that once again consists of
antibiotic-sensitive cells and a small subpopulation of persisters (67). The switch from normal
growth to persistence and vice versa is stochastic and epigenetic in nature (12). At least in
Mycobacterium tuberculosis, the regulation of persistence appears to involve noise in gene expression amplified by positive feedback (97). Persistence is a form of bet-hedging as it ensures
survival during catastrophes (56). In addition,
ANRV354-MI62-11
ARI
5 August 2008
18:1
persistence of a subpopulation of cells might indirectly benefit other cells in a population as the
growth-arrested cells do not compete for limited resources (35). Recent mathematical modeling suggests that bacterial persistence can be
regarded as a social trait and can be influenced
by kin selection (35).
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
Sporulation Bistability as a
Bet-Hedging Strategy
Recently, quantitative time-lapse microscopy
was used to generate lineage and cell fate
maps of single B. subtilis cells growing out
to a sporulating microcolony (Figure 3). The
study demonstrated that under these conditions
B. subtilis employs a bet-hedging strategy
whereby some cells sporulate and others utilize alternative metabolites to continue growth
(and can putatively pursue other survival tactics)
(104).
For individual cells the benefit of sporulation is clear; spores are resistant to various
environmental conditions and can ensure the
preservation of the clonal lineage, whereas vegetative cells could not. In the laboratory strain,
however, a significant fraction of cells do not
use the remaining energy sources for sporulation but rather delay spore formation or avoid
it. The potential advantage for these cells is
twofold. First, these cells increase in number
and may sporulate later using nutrients re-
leased by cells that have lysed. This resource
use, termed cannibalism or fratricide, has been
demonstrated in a number of studies (21 and
references therein). Second, these cells are capable of rapidly resuming growth in the event
of a new flux of nutrients. In contrast, cells that
have sporulated are committed to a long-term
process of spore formation and subsequent germination. Each of these paths is a form of specialization that increases efficiency in one area
at the expense of the other.
OUTLOOK
The strategies and mechanisms discussed in
this review are not limited to the microorganisms mentioned here. Many other bacterial
and fungal species display phenotypic variation
that may reflect a form of bet-hedging (see 11,
25, 92, 93 and references therein). These include processes that affect intra- or interspecies
competition, as well as host-pathogen interactions, such as mucoidy and cytotoxicity of Pseudomonas aeruginosa and bacteriocin production
in E. coli. A major challenge for future research
will be to assess the effects of variable phenotypes on the interactions between organisms
under steady-state and fluctuating conditions.
This finding(s) may shed light on the pressures
responsible for the evolution of genetic networks that directly or indirectly result in population multistability.
SUMMARY POINTS
1. Research increasingly acknowledges the presence and importance of cell-to-cell variability for the perpetuation of clonal populations.
2. Multistability is a ubiquitous feature of bacteria involving many different processes.
3. Phenotypic variable populations show increased fitness compared with homogeneous
populations under fluctuating environments.
4. Genetic logic-AND gates are common network motifs in bacteria to generate heterogeneity.
5. Cell states can be passed on from one generation to the next via EI and this process might
be important in bacterial development.
6. Synthetic biology and qualitative analyses of network motifs are promising for biotechnological and medical applications.
www.annualreviews.org • Bistability in Bacteria
205
ANRV354-MI62-11
ARI
5 August 2008
18:1
DISCLOSURE STATEMENT
The authors are not aware of any biases that might be perceived as affecting the objectivity of this
review.
ACKNOWLEDGMENTS
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
We apologize to those whose research could not be cited due to space limitations. JWV was
supported by an Intra-European Marie-Curie Fellowship from the European Commission, and by
a grant from the Biotechnology and Biological Sciences Research Council awarded to J. Errington.
WKS was supported by a Rubicon fellowship from the Netherlands Organization of Scientific
Research (NWO). JWV and WKS contributed equally to this manuscript.
LITERATURE CITED
2. Elegant experimental
study that shows how
switching affects
population growth.
206
1. Acar M, Becskei A, van Oudenaarden A. 2005. Enhancement of cellular memory by reducing stochastic
transitions. Nature 435:228–32
2. Acar M, Mettetal JT, van Oudenaarden A. 2008. Stochastic switching as a survival strategy in
fluctuating environments. Nat. Genet. 40:471–75
3. Ackermann M, Stearns SC, Jenal U. 2003. Senescence in a bacterium with asymmetric division. Science
300:1920
4. Aguilaniu H, Gustafsson L, Rigoulet M, Nystrom T. 2003. Asymmetric inheritance of oxidatively damaged proteins during cytokinesis. Science 299:1751–53
5. Aguilar C, Vlamakis H, Losick R, Kolter R. 2007. Thinking about Bacillus subtilis as a multicellular
organism. Curr. Opin. Microbiol. 10:638–43
6. Ajo-Franklin CM, Drubin DA, Eskin JA, Gee EP, Landgraf D, et al. 2007. Rational design of memory
in eukaryotic cells. Genes Dev. 21:2271–76
7. Alper H, Moxley J, Nevoigt E, Fink GR, Stephanopoulos G. 2006. Engineering yeast transcription
machinery for improved ethanol tolerance and production. Science 314:1565–68
8. Angeli D, Ferrell JE Jr, Sontag ED. 2004. Detection of multistability, bifurcations, and hysteresis in a
large class of biological positive-feedback systems. Proc. Natl. Acad. Sci. USA 101:1822–27
9. Artyomov MN, Das J, Kardar M, Chakraborty AK. 2007. Purely stochastic binary decisions in cell
signaling models without underlying deterministic bistabilities. Proc. Natl. Acad. Sci. USA 104:18958–63
10. Atkinson MR, Savageau MA, Myers JT, Ninfa AJ. 2003. Development of genetic circuitry exhibiting
toggle switch or oscillatory behavior in Escherichia coli. Cell 113:597–607
11. Avery SV. 2006. Microbial cell individuality and the underlying sources of heterogeneity. Nat. Rev.
Microbiol. 4:577–87
12. Balaban NQ, Merrin J, Chait R, Kowalik L, Leibler S. 2004. Bacterial persistence as a phenotypic switch.
Science 305:1622–25
13. Becskei A, Seraphin B, Serrano L. 2001. Positive feedback in eukaryotic gene networks: cell differentiation
by graded to binary response conversion. EMBO J. 20:2528–35
14. Bigger JW. 1944. Treatment of Staphylococcal infections with penicillin by intermittent sterilisation.
Lancet 244:497–500
15. Bishop AL, Rab FA, Sumner ER, Avery SV. 2007. Phenotypic heterogeneity can enhance rare-cell
survival in ‘stress-sensitive’ yeast populations. Mol. Microbiol. 63:507–20
16. Bren A, Eisenbach M. 2001. Changing the direction of flagellar rotation in bacteria by modulating the
ratio between the rotational states of the switch protein FliM. J. Mol. Biol. 312:699–709
17. Cahn FH, Fox MS. 1968. Fractionation of transformable bacteria from competent cultures of Bacillus
subtilis on renografin gradients. J. Bacteriol. 95:867–75
18. Casadesus J, D’Ari R. 2002. Memory in bacteria and phage. Bioessays 24:512–18
19. Casadesus J, Low D. 2006. Epigenetic gene regulation in the bacterial world. Microbiol. Mol. Biol. Rev.
70:830–56
Veening
· ·
Smits
Kuipers
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
ANRV354-MI62-11
ARI
5 August 2008
18:1
20. Chai Y, Chu F, Kolter R, Losick R. 2007. Bistability and biofilm formation in Bacillus subtilis. Mol.
Microbiol. 67:254–63
21. Claverys JP, Havarstein LS. 2007. Cannibalism and fratricide: mechanisms and raisons d’etre. Nat. Rev.
Microbiol. 5:219–29
22. Cohen D. 1966. Optimizing reproduction in a randomly varying environment. J. Theor. Biol. 12:119–29
23. Cohn M, Horibata K. 1959. Analysis of the differentiation and of the heterogeneity within a population
of Escherichia coli undergoing induced beta-galactosidase synthesis. J. Bacteriol. 78:613–23
24. Cox RS 3rd, Surette MG, Elowitz MB. 2007. Programming gene expression with combinatorial promoters. Mol. Syst. Biol. 3:145
25. Dubnau D, Losick R. 2006. Bistability in bacteria. Mol. Microbiol. 61:564–72
26. Dworkin J, Losick R. 2005. Developmental commitment in a bacterium. Cell 121:401–9
27. Elowitz MB, Leibler S. 2000. A synthetic oscillatory network of transcriptional regulators. Nature
403:335–38
28. Elowitz MB, Levine AJ, Siggia ED, Swain PS. 2002. Stochastic gene expression in a single cell.
Science 297:1183–86
29. Ferrell JE Jr. 2002. Self-perpetuating states in signal transduction: positive feedback, double-negative
feedback and bistability. Curr. Opin. Cell Biol. 14:140–48
30. Foster PL. 2007. Stress-induced mutagenesis in bacteria. Crit. Rev. Biochem. Mol. Biol. 42:373–97
31. Fujita M, Gonzalez-Pastor JE, Losick R. 2005. High- and low-threshold genes in the Spo0A regulon of
Bacillus subtilis. J. Bacteriol. 187:1357–68
32. Fujita M, Losick R. 2005. Evidence that entry into sporulation in Bacillus subtilis is governed by a gradual
increase in the level and activity of the master regulator Spo0A. Genes Dev. 19:2236–44
33. Galhardo RS, Hastings PJ, Rosenberg SM. 2007. Mutation as a stress response and the regulation of
evolvability. Crit. Rev. Biochem. Mol. Biol. 42:399–435
34. Garcia-Ojalvo J, Elowitz MB, Strogatz SH. 2004. Modeling a synthetic multicellular clock: repressilators
coupled by quorum sensing. Proc. Natl. Acad. Sci. USA 101:10955–60
35. Gardner A, West SA, Griffin AS. 2007. Is bacterial persistence a social trait? PLoS ONE 2:e752
36. Gardner TS, Cantor CR, Collins JJ. 2000. Construction of a genetic toggle switch in Escherichia coli.
Nature 403:339–42
37. Gardner TS, Collins JJ. 2005. Bistable genetic toggle switch. U.S. Patent No. 6841376
38. Gerdes K, Christensen SK, Lobner-Olesen A. 2005. Prokaryotic toxin-antitoxin stress response loci.
Nat. Rev. Microbiol. 3:371–82
39. Godoy VG, Gizatullin FS, Fox MS. 2000. Some features of the mutability of bacteria during nonlethal
selection. Genetics 154:49–59
40. Goldstein J, Pollitt NS, Inouye M. 1990. Major cold shock protein of Escherichia coli. Proc. Natl. Acad.
Sci. USA 87:283–87
41. Hadden C, Nester EW. 1968. Purification of competent cells in the Bacillus subtilis transformation system.
J. Bacteriol. 95:876–85
42. Haijema BJ, Hahn J, Haynes J, Dubnau D. 2001. A ComGA-dependent checkpoint limits growth during
the escape from competence. Mol. Microbiol. 40:52–64
43. Hammer K, Mijakovic I, Jensen PR. 2006. Synthetic promoter libraries—tuning of gene expression.
Trends Biotechnol. 24:53–55
44. Hamoen LW, Venema G, Kuipers OP. 2003. Controlling competence in Bacillus subtilis: shared use of
regulators. Microbiology 149:9–17
45. Hasty J, McMillen D, Collins JJ. 2002. Engineered gene circuits. Nature 420:224–30
46. Heinemann M, Panke S. 2006. Synthetic biology—putting engineering into biology. Bioinformatics
22:2790–99
47. Ireton K, Rudner DZ, Siranosian KJ, Grossman AD. 1993. Integration of multiple developmental signals
in Bacillus subtilis through the Spo0A transcription factor. Genes Dev. 7:283–94
48. Jablonka E, Lamb MJ. 2006. Evolution in Four Dimensions: Genetic, Epigenetic, Behavioral, and
Symbolic Variation in the History of Life. Cambridge, MA: MIT Press
49. Kærn M, Elston TC, Blake WJ, Collins JJ. 2005. Stochasticity in gene expression: from theories to
phenotypes. Nat. Rev. Genet. 6:451–64
www.annualreviews.org • Bistability in Bacteria
28. Together with
Reference 78, this study
pioneered the use of an
in vivo method to
analyze and quantify
noise.
48. Thought-provoking
book in which the
authors argue that there
is more to heredity than
genes.
207
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
ANRV354-MI62-11
ARI
5 August 2008
62. Experimental study
showing directly for the
first time that noise in
comK transcription
determines entry into
the competent state in
B. subtilis.
208
18:1
50. Karmakar R, Bose I. 2007. Positive feedback, stochasticity and genetic competence. Phys. Biol. 4:29–37
51. Kaufmann BB, Yang Q, Mettetal JT, van Oudenaarden A. 2007. Heritable stochastic switching revealed
by single-cell genealogy. PLoS Biol. 5:e239
52. Kearns DB, Chu F, Branda SS, Kolter R, Losick R. 2005. A master regulator for biofilm formation by
Bacillus subtilis. Mol. Microbiol. 55:739–49
53. Kobayashi K. 2007. Gradual activation of the response regulator DegU controls serial expression of
genes for flagellum formation and biofilm formation in Bacillus subtilis. Mol. Microbiol. 66:395–409
54. Korobkova E, Emonet T, Vilar JM, Shimizu TS, Cluzel P. 2004. From molecular noise to behavioural
variability in a single bacterium. Nature 428:574–78
55. Krishnamurthy S, Smith E, Krakauer D, Fontana W. 2007. The stochastic behavior of a molecular
switching circuit with feedback. Biol. Direct. 2:13
56. Kussell E, Leibler S. 2005. Phenotypic diversity, population growth, and information in fluctuating
environments. Science 309:2075–78
57. Leisner M, Stingl K, Radler JO, Maier B. 2007. Basal expression rate of comK sets a ‘switching-window’
into the K-state of Bacillus subtilis. Mol. Microbiol. 63:1806–16
58. Lewis K. 2007. Persister cells, dormancy and infectious disease. Nat. Rev. Microbiol. 5:48–56
59. Lindner AB, Madden R, Demarez A, Stewart EJ, Taddei F. 2008. Asymmetric segregation of protein
aggregates is associated with cellular aging and rejuvenation. Proc. Natl. Acad. Sci. USA. 105:3076–81
60. Lisman JE. 1985. A mechanism for memory storage insensitive to molecular turnover: a bistable autophosphorylating kinase. Proc. Natl. Acad. Sci. USA 82:3055–57
61. Maamar H, Dubnau D. 2005. Bistability in the Bacillus subtilis K-state (competence) system requires a
positive feedback loop. Mol. Microbiol. 56:615–24
62. Maamar H, Raj A, Dubnau D. 2007. Noise in gene expression determines cell fate in Bacillus
subtilis. Science 317:526–29
63. Maynard Smith J. 1964. Group selection and kin selection. Nature 201:1145–47
64. Mayo AE, Setty Y, Shavit S, Zaslaver A, Alon U. 2006. Plasticity of the cis-regulatory input function of
a gene. PLoS. Biol. 4:e45
65. Molle V, Fujita M, Jensen ST, Eichenberger P, Gonzalez-Pastor JE, et al. 2003. The Spo0A regulon of
Bacillus subtilis. Mol. Microbiol. 50:1683–701
66. Monod J, Jacob F. 1961. Teleonomic mechanisms in cellular metabolism, growth, and differentiation.
Cold Spring Harb. Symp. Quant. Biol. 26:389–401
67. Moyed HS, Broderick SH. 1986. Molecular cloning and expression of hipA, a gene of Escherichia coli
K-12 that affects frequency of persistence after inhibition of murein synthesis. J. Bacteriol. 166:399–403
68. Msadek T. 1999. When the going gets tough: survival strategies and environmental signaling networks
in Bacillus subtilis. Trends Microbiol. 7:201–7
69. Mukai K, Kawata M, Tanaka T. 1990. Isolation and phosphorylation of the Bacillus subtilis degS and degU
gene products. J. Biol. Chem. 265:20000–6
70. Muller-Hill B. 1996. The lac Operon: A Short History of a Genetic Paradigm. Berlin: Walter de Gruyter
71. Murphy KF, Balazsi G, Collins JJ. 2007. Combinatorial promoter design for engineering noisy gene
expression. Proc. Natl. Acad. Sci. USA 104:12726–31
72. Neildez-Nguyen TM, Parisot A, Vignal C, Rameau P, Stockholm D, et al. 2008. Epigenetic gene
expression noise and phenotypic diversification of clonal cell populations. Differentiation 76:33–40
73. Nester EW, Stocker BA. 1963. Biosynthetic latency in early stages of deoxyribonucleic acid transformation in Bacillus subtilis. J. Bacteriol. 86:785–96
74. Ninfa AJ, Mayo AE. 2004. Hysteresis vs graded responses: the connections make all the difference. Sci.
STKE 2004:e20
75. Ninfa AJ, Selinsky S, Perry N, Atkins S, Xiu SQ, et al. 2007. Using two-component systems and other
bacterial regulatory factors for the fabrication of synthetic genetic devices. Methods Enzymol. 422:488–512
76. Novick A, Weiner M. 1957. Enzyme induction as an all-or-none phenomenon. Proc. Natl. Acad. Sci. USA
43:553–66
77. Ortega F, Garces JL, Mas F, Kholodenko BN, Cascante M. 2006. Bistability from double phosphorylation in signal transduction. Kinetic and structural requirements. FEBS J. 273:3915–26
Veening
· ·
Smits
Kuipers
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
ANRV354-MI62-11
ARI
5 August 2008
18:1
78. Ozbudak EM, Thattai M, Kurtser I, Grossman AD, van Oudenaarden A. 2002. Regulation of
noise in the expression of a single gene. Nat. Genet. 31:69–73
79. Ozbudak EM, Thattai M, Lim HN, Shraiman BI, van Oudenaarden A. 2004. Multistability in the lactose
utilization network of Escherichia coli. Nature 427:737–40
80. Partridge L, Barton NH. 1993. Optimality, mutation and the evolution of ageing. Nature 362:305–11
81. Paulsson J. 2004. Summing up the noise in gene networks. Nature 427:415–18
82. Pennington JM, Rosenberg SM. 2007. Spontaneous DNA breakage in single living Escherichia coli cells.
Nat. Genet. 39:797–802
83. Ponder RG, Fonville NC, Rosenberg SM. 2005. A switch from high-fidelity to error-prone DNA doublestrand break repair underlies stress-induced mutation. Mol. Cell 19:791–804
84. Rando OJ, Verstrepen KJ. 2007. Timescales of genetic and epigenetic inheritance. Cell 128:655–68
85. Raser JM, O’Shea EK. 2005. Noise in gene expression: origins, consequences, and control. Science
309:2010–13
86. Renzette N, Gumlaw N, Nordman JT, Krieger M, Yeh SP, et al. 2005. Localization of RecA in Escherichia
coli K-12 using RecA-GFP. Mol. Microbiol. 57:1074–85
87. Rosche WA, Foster PL. 1999. The role of transient hypermutators in adaptive mutation in Escherichia
coli. Proc. Natl. Acad. Sci. USA 96:6862–67
88. Simmons LA, Grossman AD, Walker GC. 2007. Replication is required for the RecA localization response to DNA damage in Bacillus subtilis. Proc. Natl. Acad. Sci. USA 104:1360–65
89. Smith MC, Sumner ER, Avery SV. 2007. Glutathione and Gts1p drive beneficial variability in the
cadmium resistances of individual yeast cells. Mol. Microbiol. 66:699–712
90. Smits WK, Eschevins CC, Susanna KA, Bron S, Kuipers OP, Hamoen LW. 2005. Stripping Bacillus: ComK auto-stimulation is responsible for the bistable response in competence development. Mol.
Microbiol. 56:604–14
91. Smits WK, Hoa TT, Hamoen LW, Kuipers OP, Dubnau D. 2007. Antirepression as a second mechanism
of transcriptional activation by a minor groove binding protein. Mol. Microbiol. 64:368–81
92. Smits WK, Kuipers OP, Veening JW. 2006. Phenotypic variation in bacteria: the role of feedback
regulation. Nat. Rev. Microbiol. 4:259–71
93. Smits WK, Veening JW, Kuipers OP. 2008. Phenotypic variation and bistable switching in bacteria. In
Bacterial Physiology: A Molecular Approach, ed. W El-Sharoud, pp. 339–65. Berlin/Heidelberg: SpringerVerlag
94. Stewart EJ, Madden R, Paul G, Taddei F. 2005. Aging and death in an organism that reproduces
by morphologically symmetric division. PLoS Biol. 3:e45
95. Süel GM, Garcia-Ojalvo J, Liberman LM, Elowitz MB. 2006. An excitable gene regulatory circuit
induces transient cellular differentiation. Nature 440:545–50
96. Süel GM, Kulkarni RP, Dworkin J, Garcia-Ojalvo J, Elowitz MB. 2007. Tunability and noise dependence
in differentiation dynamics. Science 315:1716–19
97. Sureka K, Ghosh B, Dasgupta A, Basu J, Kundu M, Bose I. 2008. Positive feedback and noise activate
the stringent response regulator rel in mycobacteria. PLoS ONE 3:e1771
98. Swain PS, Elowitz MB, Siggia ED. 2002. Intrinsic and extrinsic contributions to stochasticity in gene
expression. Proc. Natl. Acad. Sci. USA 99:12795–800
99. Thattai M, van Oudenaarden A. 2004. Stochastic gene expression in fluctuating environments. Genetics
167:523–30
100. Torkelson J, Harris RS, Lombardo MJ, Nagendran J, Thulin C, Rosenberg SM. 1997. Genome-wide
hypermutation in a subpopulation of stationary-phase cells underlies recombination-dependent adaptive
mutation. EMBO J. 16:3303–11
101. Veening JW, Hamoen LW, Kuipers OP. 2005. Phosphatases modulate the bistable sporulation gene
expression pattern in Bacillus subtilis. Mol. Microbiol. 56:1481–94
102. Veening JW, Igoshin OA, Eijlander RT, Nijland R, Hamoen LW, Kuipers OP. 2008. Transient heterogeneity in extracellular protease production by Bacillus subtilis. Mol. Syst. Biol. 4:184
103. Veening JW, Smits WK, Hamoen LW, Kuipers OP. 2006. Single cell analysis of gene expression
patterns of competence development and initiation of sporulation in Bacillus subtilis grown on chemically defined media. J. Appl. Microbiol. 101:531–41
www.annualreviews.org • Bistability in Bacteria
78. Together with
Reference 28, this study
pioneered the use of an
in vivo method to
analyze and quantify
noise.
94. Shows that even
symmetrical dividing
bacteria suffer from
aging.
95. Provides evidence
for the excitable model
that dictates
competence
development.
209
ANRV354-MI62-11
ARI
5 August 2008
104. Veening JW, Stewart EJ, Berngruber TW, Taddei F, Kuipers OP, Hamoen LW. 2008. Bethedging and epigenetic inheritance in bacterial cell development. Proc. Natl. Acad. Sci. USA
105:4393–98
105. Verhamme DT, Kiley TB, Stanley-Wall NR. 2007. DegU co-ordinates multicellular behaviour exhibited
by Bacillus subtilis. Mol. Microbiol. 65:554–68
106. Vlamakis H, Aguilar C, Losick R, Kolter R. 2008. Control of cell fate by the formation of an architecturally complex bacterial community. Genes Dev. 22:945–53
107. Yildirim N, Santillan M, Horike D, Mackey MC. 2004. Dynamics and bistability in a reduced model of
the lac operon. Chaos 14:279–92
108. Zambrano MM, Kolter R. 1996. GASPing for life in stationary phase. Cell 86:181–84
Annu. Rev. Microbiol. 2008.62:193-210. Downloaded from arjournals.annualreviews.org
by University of Groningen on 10/31/08. For personal use only.
104. Time-lapse study
that shows that
sporulation in B. subtilis
is a bet-hedging strategy
and that the signal to
sporulate can be
epigenetically inherited.
18:1
210
Veening
· ·
Smits
Kuipers